Patients and methods

PATIENTS

We investigated two groups of patients with HAB only in the central nervous system. Group 1 consisted of 61 UK and 23 Dutch HAB patients with a single HAB. These cases were ascertained with the help of neurosurgeons, neurologists, internists, and clinical geneticists. In addition to DNA analysis, all patients underwent clinical examinations for detection of VHL associated tumours (ophthalmological examination and abdominal sonography or MRI) with negative findings. Group 2 consisted of four patients with multiple HABs, but no other evidence of VHL disease (that is, absence of further VHL related tumours) on clinical screening and radiological screening. All patients had histopathologically confirmed HAB at operation.

AGE AT DIAGNOSIS

Between 1973 and 1996 a total of 182 HAB patients were reported to the National Dutch Pathological Archive (Palga). Fig 1 shows the age at diagnosis of Palga patients as well as patients studied. Details of age distribution of a previous population based cohort of UK HAB patients have been reported previously.27

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Number of patients and age at diagnosis for 182 patients with haemangioblastoma (HAB) in the central nervous system (CNS) reported in the National Dutch Pathological Archive (Palga) between 1973 and 1996 and 26 Dutch patients in this study with a single HAB between 1996 and 1999.

In 1996, guidelines were distributed via the Dutch newsletter for neurologists on screening all patients with a HAB for a VHL germline mutation. The Dutch patients in the present study were referred for DNA diagnosis from 1996 to 1999, and the mean age at diagnosis was 37.5 years (range 14-71 years). Compared to unselected cases, the age distribution in the UK as well as in the Dutch cases in this study was biased towards an earlier age at onset.

DNA ANALYSIS

High molecular weight DNA of the probands was isolated from peripheral blood according to established procedures. Exons 1, 2, and 3 of the VHL gene and their immediately flanking sequences were amplified using the polymerase chain reaction.29 The flanking sequences included 90 nucleotides upstream of the start codon (the first nucleotide of the coding region is 214) and 45 nucleotides downstream of the stop codon. The nucleotides are numbered according to Latif et al 15 (Genbank accession number L15409). PCR products were purified and subjected to sequence analysis using either an ABI automated sequencer or the dideoxy chain termination reaction with a pUC sequencing kit (Boehringer Mannheim, Mannheim, Germany), using γ-³⁵S dATP (600 Ci/mmol). The amplification primers were used as primers in the sequencing reactions. In the UK cases, intragenic mutations were also sought by single strand conformation polymorphism (SSCP).

Screening for genetic rearrangements and deletions was performed by Southern blot analysis, or a novel PCR based deletion assay (Dowet al, in preparation). In Southern blot analysis, DNA was digested with EcoRI alone or with EcoRI andAseI double digest. After gel electrophoresis and transfer to Hybond-N filters, the genomic DNA was hybridised with the VHL g7-cDNA probe,15 according to the manufacturer's instruction. Quantitative Southern blotting was performed by hybridising genomic DNA with the VHL g7-cDNA probe and with a beta globin probe, to detect deletions encompassing the entire VHL gene.30

Results

SINGLE HAB

Eighty four cases with single HAB in the central nervous system (23 Dutch, 61 UK) were investigated. The age distribution of these cases is shown in fig 2. VHL germline mutations were identified in three cases (table 1). The incidence of aVHL germline mutation in single HAB cases younger than 30 years was 4.3% (1/23), 30-39 years was 4.5% (1/22), 40-49 years was 4.2% (1/24), and 50 years or older was 0% (0/15).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Distribution of single HAB patients (dark = UK, light = Dutch) by age at diagnosis.

Details of the three germline mutations identified were as follows. (1) A C to T transition at nucleotide 454 was detected in a 44 year old woman (D24) with a single cerebellar HAB (fig 3). This missense mutation leads to a change of proline to serine at codon 81 (P81S) in the VHL protein. DNA analysis of other close relatives showed that four clinically unaffected first and second degree relatives (aged 17-77 years) were also carriers of a P81S germline mutation. (2) A 7 bp frameshift mutation (del 582 GACACAC) was detected in a 29 year old patient (B1) with a single cerebellar HAB with no family history and no evidence of VHL disease on clinical and radiological screening. However, she subsequently developed pancreatic cysts at 34 years of age. (3) A large germline deletion was identified by Southern blot analysis in a 36 year old woman (B2) with a single cerebellar HAB but no other features of VHL disease.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Pedigree of patient D24, showing four clinically unaffected relatives and their ages with the VHL germline mutation P81S.

Two patients with a single HAB developed some additional features of VHL disease during the study: one patient (B1) with a VHL gene mutation (del 582 GACACAC) developed pancreatic cysts (see above) and one patient (B3) developed renal cell carcinoma aged 44 years following a cerebellar HAB at the age of 40 years, but no VHL mutation was identified.

MULTIPLE HABS

Four patients with multiple HABs (two Dutch and two UK) and without additional VHL related tumours were analysed for VHL germline mutations. Deletions were detected in two patients, (1) a male (B5) with one cerebellar and one medullary HAB at the age of 40 years, and (2) a male (B6) with multiple spinal HABs at the age of 44 years. Germline mutations were not identified in the two Dutch patients (D13 and D32) with both cerebellar and spinal HABs (aged 44 and 66 years).

Discussion

We found that the overall risk for finding a VHL germline mutation in a population of 84 patients with a single HAB in the central nervous system and no further features of VHL disease at the time of diagnosis was approximately 4%. In clinical studies of sporadic patients with a HAB it was suggested that a substantial proportion of HAB could be associated with VHL disease upon more detailed examination, that is, 23% to 34.3% were found to be affected with VHL disease.6 28 Molecular genetic analysis of the VHL gene had indicated that sporadic patients with a HAB may have a risk of a VHL gene germline mutation of approximately 10%.31 32 In our larger study, we found a lower detection rate despite using more sensitive methods of VHL gene analysis, but this is presumably related to the very careful clinical and radiological screening performed before entry into our study. The identification of a VHL germline mutation has important implications for the risk of further tumours and for the risk of VHL disease in relatives, so the identification of a VHL mutation in a significant proportion of single HAB patients despite conventional clinical and radiological investigation is important.

Statistical analysis of the age at onset of HAB in VHL disease and non-VHL cases (based on clinical criteria) is consistent with a one and two hit tumourigenesis model as predicted by the Knudson hypothesis.27 Mean age at diagnosis of cerebellar HAB in VHL disease is younger than in sporadic cases (29 years versus 48 years, respectively) so we anticipated a higher incidence of unsuspected VHL gene mutations in early onset cases. In addition, older patients with VHL gene mutations would be more likely to manifest other evidence of VHL disease and so be excluded from our study. Thus, although our results were broadly compatible with this hypothesis, the identification of germline VHL gene mutations in two of a group of 46 patients aged 30 to 49 years with a single HAB and no clinical or radiological features of VHL disease suggests that molecular genetic analysis should be used in all single HAB patients younger than 50 years. For older onset patients, the frequency of germline mutations is likely to be less (see above) and larger research based studies are required to define whether molecular genetic analysis should be performed when clinical and radiological screening has shown no evidence of VHL disease.

A major strength of the present study was the use of recently developed techniques to detect large germline deletions. Before the introduction of these techniques, there was a VHLgermline mutation detection rate (with Southern blotting and sequencing of the coding region) of approximately 80% in known VHL families.22 However, methods to detect large deletions (for example, quantitative Southern blotting) have significantly increased the detection rate, reaching 100% in proven familial VHL disease.30 Therefore a conservative estimate of the mutation detection sensitivity of the strategy used in this study would be in the order of 95%. Although other studies of sporadic patients with a HAB have used less sensitive techniques,31-33 it is of interest that Oberstrass et al 31 detected a germline mutation in two of 20 patients (aged 18 and 40 years) with HAB of the central nervous system (although no data were available about a possible family history of VHL disease). Decker et al 33reported a 29 year old patient with recurrent spinal HAB, a negative family history of VHL disease, and a de novo frameshiftVHL gene exon 2 mutation. However, this case was not comparable to those in our study because a renal mass and pancreatic and renal cysts were detected on clinical screening. In a series of 18 sporadic patients with a HAB, Olschwanget al 32 found a missense mutation in two patients (42 and 56 years of age) without clinical investigations showing any evidence of VHL disease. The latter case would only have been detected by also screening single HAB patients aged over 50 years for a VHL gene mutation.

Although specific missense VHL mutations may cause a phaeochromocytoma only phenotype,24-26 we did not find unequivocal evidence for VHL gene mutations that would predispose to a HAB only phenotype. However, the VHLgermline mutation (P81S) that was found in a 44 year old woman with a solitary HAB and in four clinically unaffected family members was associated with an unusually low penetrance within this family. To reduce the possibility of a genetic polymorphism, 50 non-VHL patients were sequenced. Sequence analysis of exon 1 showed that all persons were homozygous for nucleotide 545C (data not shown). Phenotypic expression in VHL disease is influenced by allelic heterogeneity, stochastic events, and genetic modifiers.34 The P81S mutation has been reported four times previously: (1) in an isolated German patient with a full blown VHL tumour spectrum (cerebellar and spinal HAB, renal cell carcinoma, and renal, pancreatic, and epididymal cysts); (2) in a 34 year old American patient with HAB only; (3) in a 35 year old American patient with retinal haemangioblastoma and islet cell tumour of the pancreas (the father is the only other relative with a VHL related tumour and had a phaeochromocytoma); and (4) in an isolated Japanese patient with multiple HABs and a renal cell carcinoma.21 30 35 These findings suggest that P81S mutation carriers in the family are also at risk of renal cell carcinoma. Moreover, only one of the P81S carriers had affected family members, which may imply that this missense mutation has a low penetrance.

Interestingly, we did not find VHL germline mutations in two of the four patients with multiple HABs. As the presence of two or more retinal or cerebellar HAB satisfies the strict diagnostic criteria for VHL disease,12 14 this was an unexpected finding in the light of the high sensitivity of the mutation detection methods used, and could perhaps indicate additional HAB susceptibility gene(s). Alternative explanations would include a mutation in part of the VHL gene not analysed (for example, regulatory domain) or somatic mosaicism. In this context it is interesting that the two multiple HAB patients with germline mutations had the earliest age at onset and the patients with later onset may be mosaic and so have a milder phenotype or represent phenocopies (independent mutation events giving rise to various HAB could also be expected by chance). Although mosaicism has so far been described infrequently in VHL disease,36 it seems likely that it will be relatively common, as in neurofibromatosis type 2.37

We have shown that VHL gene mutation analysis facilitates the management of sporadic patients with HAB and recommend that it should be performed in all patients younger than 50 years with a single HAB even if there is no other clinical or radiological evidence of VHL disease. Although the detection rate in older patients should be lower, we suggest that such patients need to be studied on a research basis, using the latest mutation detection strategies to define cost-benefit consequences for molecular genetic analysis of this group of patients.